Free calculator for GPT, Claude, Gemini, DeepSeek and 500+ other LLMs. Paste your prompt or enter token counts to see the exact cost across every model on the market. Live data — updated daily from OpenRouter pricing. No signup, no ads on the calculator itself.
Live cost for each major model at common prompt sizes. Numbers update with current OpenRouter pricing.
| Prompt type | Tokens (in / out) | OpenAI | DeepSeek | Anthropic | |
|---|---|---|---|---|---|
| Short Q&A | 500 / 200 | $0.00 | $0.000041 | $0.0002 | $0.0004 |
| Code review (1K) | 1,000 / 500 | $0.00 | $0.000095 | $0.0005 | $0.0009 |
| Long doc summarization | 8,000 / 1,000 | $0.00 | $0.0004 | $0.0026 | $0.0033 |
| Agent task (typical) | 4,000 / 2,000 | $0.00 | $0.0004 | $0.0019 | $0.0035 |
| Heavy agent run | 50,000 / 10,000 | $0.00 | $0.0028 | $0.017 | $0.025 |
| 1M in / 1M out (scale) | 1,000,000 / 1,000,000 | $0.00 | $0.160 | $0.669 | $1.50 |
Assumes each model's flagship tier (GPT-5.4, Claude Opus 4.6, Gemini 2.5 Pro, DeepSeek V3). Switch to cheaper variants like nano or mini in the table below to see dramatic cost drops (often 10–30×).
Lowest-cost models with non-zero pricing, sorted by blended (3:1) input/output price. Use the calculator below to compare for your exact workload.
Cost = (input_tokens × input_price + output_tokens × output_price) ÷ 1,000,000. Both prices are in USD per 1M tokens. We pull live pricing from OpenRouter's public API, which mirrors what providers charge.
Each provider sets its own rates based on model capability, compute cost, and competitive positioning. Premium reasoning models (Claude Opus, GPT high-effort) cost more than fast variants (nano, mini, flash) because they spend more compute per token.
For most workloads, look at the fast tier: Claude Haiku, GPT-5 mini, Gemini Flash, DeepSeek V3. They're typically 10–30× cheaper than flagship models while staying good enough for chat, classification, and routing. Reserve flagship models for reasoning-heavy tasks.
No. This shows list prices. Most providers offer prompt caching (50–90% off cached input) and batch discounts (50% off async jobs). Check each provider's docs for the exact numbers.
Found the model that fits your budget? Sign up via one of these providers to start using it. Some links earn us a small commission at no extra cost to you.
See our affiliate disclosure for the full list and how we choose what to recommend.